Artificial Intelligence, Machine Learning and Automated Service Assurance: Telecommunications Products Every Operator Needs to Understand.
Once the preserve of (often terrifying) science-fiction films and stories, the idea that computers can “think” has become a reality. Artificial intelligence (AI) and machine learning (ML) touch almost every aspect of 20th-century life and, unlike in Hollywood’s vision, are benign. What’s more, they are so unobtrusive that we barely even notice them in practice; for example, when was the last time you considered how your movie-streaming provider was able to make such accurate suggestions for shows you would like, or why the automatic vacuum cleaner always manages to leave the kitchen floor spotless? There’s AI at work in both scenarios – with the machines you use “learning” through fairly simple algorithms about your preferences and the layout of your home.
And that’s just domestic, day to day life. In some cases, modern industries would be hard-pressed to operate at all without the assistance of AI and ML. The telecommunications sector is probably the best example, as it is here that the pace and range of other technological changes (think of the huge demands of 5G and the Internet of Things) have forced evolution in the way processes such as assurance are carried out.
For telcos, there are now too many elements in play for assurance to be guaranteed or delivered manually. In the next five years or so, there will be more than 20 billion devices communicating with each other via telecommunications networks. Operators need telecommunications products that can automate the assurance processes involved in such a complex web – and these products must combine high-security with flawless delivery if customers are not to migrate to the competition.
There are three key ways in which AI and ML come into play here. All of them rely on intelligent assurance and automated analysis to some extent, demanding human intervention only when required. And the goal of all is to keep systems and services running without any glitches, to the extent that problems are solved before they become an issue for customers.
On one level, it’s possible to automate the detection, analysis and removal of network issues. This process of automated situation generation is generally deployed in network and service monitoring and operational centers; its strength is that, even if an issue leaks through to the customer level, AI can still take over the resolution process with no human interaction, to find the root-cause of which several events could be the symptoms and keep the network open and operational.
Then there is automated problem generation, which again requires no manual actions (and therefore no human error) if deployed correctly. In this process, machine learning tools are deployed to pinpoint the relationship between issues and remove them.
For some issues, a degree of human interaction is required – for example in the case of automated baseline generation and anomaly detection. In this situation, the starting point for AI analysis is the information about parameters and norms provided by human operators. The machine’s job is to find out where these norms are transgressed, and why.
Assurance automation in each situation is either high or total, helping telcos to maintain the level of customer experience through one of several vital AI-based telecommunications products.